Congestion Pricing for Infrastructure: I Still Don’t Know Why Net Neutrality is Important

A month ago I wrote a blog post questioning the importance of net neutrality: Remind Me: Why Should I Care about Net Neutrality? Luckily for me, law professor seeks to answer my question in a short piece in Wired, Why you should care about net neutrality. Sadly, I’m not convinced.

Frischmann (and co-author argue that the Internet should be a free for all, just as roads supposedly are, or used to be. While both roads and the Internet have to deal with the scourge of congestion, Frischmann believes it must be resolved by waiting (except in the case of emergency vehicles, of course). If the sole protocol is the first-come, first-served rule, the pain will be spread fairly:

But while smart systems seem attractive, they’ll inevitably be optimised for corporate profit and control. The principle of first-come, first-served is our best protection against interference. We need it on the web – and on the roads.

Frischmann is regarded by his critics as a bit of an ideologue, wedded to the open access model of shared infrastructure and unwilling to consider issues raised by public choice theory. A review of Frischmann’s book Infrastructure: The Social Value of Shared Resources by Adam Thierer highlights shortcomings in his open access model.

Such a Quaint Story

For starters, Frischmann’s view of roadway management is out of date. Economists have long realized that indiscriminate approaches to traffic management on the roads don’t lead to welfare-maximizing results. The high-tech alternative is congestion pricing, a system that encourages drivers to trade money for time and vice versa.

Congestion pricing wasn’t practical to implement in the past, but advanced technology has changed all that. Singapore is on its second generation of congestion pricing now, with a system that sets dynamic prices for access to congested areas:

In 1998, Singapore replaced the system with the Electronic Road Pricing (ERP) program, which uses modern technology. At the start of the journey a Cash Card is inserted into the On-Board Unit (OBU), which is fixed permanently in the vehicle and powered by the vehicle battery. When passing an ERP gantry the cash balance after the ERP charge deduction is shown on the OBU for 10 seconds. The electronic system has the ability to vary the prices based on traffic conditions and by vehicle type, time and location. Today all vehicles are charged, only emergency vehicles are exempted. In 2005 the coverage of ERP expanded the gantries around Singapore central business district and on major arterials and expressways. To ensure optimal use of road space and to maintain optimal speeds, the system is revised quarterly.

This system measures congestion dynamically and uses AI and big data to set prices optimally. The net result of this system is greater use of public transit and less congestion. This leads to lower carbon emissions, of course.

Modern Approaches to Congestion Pricing

An op-ed in the Wall Street Journal by economists Peter Cramton and R. Richard Geddes, “How Technology Can Eliminate Traffic Congestion“, examines the benefits of congestion pricing on the level of nations. One of the benefits of dynamic pricing systems they cite is smarter investments in roadway upgrades:

Accurate road prices will also help us make smarter infrastructure investments. A new lane, for instance, can be targeted to where its value—as reflected in prices—is greatest. Real-time road prices will reveal that value, thus reducing or eliminating the politicization that has afflicted infrastructure investment.

Of course, we have dynamic road pricing in the US as well: When I drive to Boulder, Colorado on US 36, I can choose a fast lane over lanes with the normal level of congestion. Drivers in the San Francisco Bay Area have the same option, one that I used on several occasions when running late to a meeting or a flight.

What Does this Have to Do with the Internet?

Arguing that the Internet should work just like the roads actually means that “pay for priority” fast lanes should be legal. Frischmann denied this in a Twitter exchange:

In fact, access to restricted areas is an example of paying for priority as the alternative is taking public transit. Prices are set by area, which effectively means destination. The whole system is based on willingness to pay (WTP). The benefit that Singapore cites is increased use of public transit.

We Can’t Simply Build our Way out of Congestion

Congestion is a fact of life in shared infrastructures that we can’t eliminate by adding capacity willy-nilly. The trouble with adding capacity is that people take it as an invitation to use the commons more intensely, so the congestion comes back.

A congestion-free infrastructure would need to be orders of magnitude more expensive today’s systems. If such an infrastructure did solve the congestion problem, it would ultimately succeed because high prices reduce demand.

So the way out of the congested infrastructure dilemma is to put supply and demand into equilibrium. This generally requires pricing access in a way that promotes investment. It also means putting buyers and sellers into a dialog with each other, something that doesn’t happen when regulators erect walls between us.

This latter dynamic underscores the importance of Thierer’s shoutout to public choice theory. Because regulators are no smarter or more virtuous than the rest of us, agencies will always be subject to capture and special interest manipulation.

Quality Pricing

In a recent essay, “Broadband service quality: Rationing or markets?”, network theorist (and High Tech Forum contributor) Martin Geddes argues that we don’t price and sell residential broadband service in a rational way:

Today we characterize broadband services by their bearer (e.g. cable, DSL, 4G) and service “speed”. This does not sufficiently describe the service. Specifically, measures of average or peak “speed” do not define the quality on offer, and it is the quality that determines fitness for purpose in use.

We mistakenly confuse speed with quality, and tend to regulate practices that affect speed whether they impact quality or not. Paid prioritization is the best example. Despite activist rhetoric, it’s actually possible by increase the quality of some Internet streams without affecting others.

The BITAG Differentiated Treatment of Internet Traffic report proves this claim:

When differentiated treatment is applied with an awareness of the requirements for different types of traffic, it becomes possible to create a benefit without an offsetting loss. For example, some differentiation techniques improve the performance or quality of experience (QoE) for particular applications or classes of applications without negatively impacting the QoE for other applications or classes of applications. The use and development of these techniques has value.
Geddes points out that each application has a quality floor, a threshold level of latency and loss that bounds its ability to function. We care less about peak bandwidth than about the minimum quality across a transaction, an application-level activity with a beginning and an end. But regulation still focuses on peak speeds.

Commercial Internet Pricing is Quality Based

When businesses buy Internet transit services, their contracts tend to specify volume, delay, and loss rather than speed. This is the norm because these are the factors that affect the costs of service provision.

These factors also define service quality. Once the provider promises to deliver a volume of traffic at set levels of delay and loss, connection speed becomes one of many tools to accomplish the goal.

When usage, delay tolerance, and loss tolerance are all unknowns, we fall to an unknown level of quality. While this simplifies billing, it doesn’t do justice to the needs of applications, innovation, or investment.

A side effect of switching from the current billing model to a quality-based model is that the unproductive net neutrality debate summarily ends. When users have control over the end-to-end quality of each application transaction, the means used by the provider to deliver the desired quality are unimportant.

Despite Professor Frischmann’s intervention, I still don’t think net neutrality is important.

[Disclosure: As a member of Singapore’s Economist and Regulatory Advisory Board, I have done paid consulting work for the Republic of Singapore on multiple occasions.]